[kronolith] Excessive memory usage with Kronolith

[Kevin Myer]

Hello,

Doing some additional troubleshooting to isolate some performance issues with
Kronolith, that started with a calendar import killing my directory server, and
now phase 2 continues in trying to actually view imported events :)

I exported my calendar from my production Horde install (not quite up-to-date,
but imports cleanly into Apple's iCal, for whatever that is worth).  I then
imported the calendar into a HEAD checkout of Kronolith (fresh).  The import
took about ten seconds, with none of the adverse affects of the previous days.

However, now when I attempt to login to my HEAD install, the loading of the
Kronolith portal block is bringing Apache to its knees.  Memory usage climbs
and climbs, until it hits my preconfigured limit (2Gb) or I manually HUP the
process.

I did some profiling with apd and here's what I found, for a trace of the
Kronolith summary Block (this is up to the point that I HUPped the process):

300,000 calls to in_array
300,000 calls to is_a
200,000 calls to horde_date:comparedate
100,000 calls to mktime, is_array, horde_date::horde_date,
horde_date::timestamp, Kronolith_event_sql::isallday,
kronolith_event_sql::getid, horde_date::comparedatetime

I have a total of 546 events in my calendar that I imported.

So, any pointers on where to look further?  I did wipe the entire database for
this install and started from scratch.  Its only when I try to access the
calendar that unhappy things start to happen.  Also tried using just the past
month (only 20 events) - same problem.  So, is it possible I have a bad event
that is throwing things off?

Kevin

-- 
Kevin M. Myer
Senior Systems Administrator
Lancaster-Lebanon Intermediate Unit 13  http://www.iu13.org